Notice: Wiley Online Library will be unavailable on Saturday 30th July 2016 from 08:00-11:00 BST / 03:00-06:00 EST / 15:00-18:00 SGT for essential maintenance. Apologies for the inconvenience.
C Wloch, Department of Healthcare Associated Infection and Antimicrobial Resistance, Health Protection Agency, 61 Colindale Avenue, London NW9 5EQ, UK. Email email@example.com
Please cite this paper as: Wloch C, Wilson J, Lamagni T, Harrington P, Charlett A, Sheridan E. Risk factors for surgical site infection following caesarean section in England: results from a multicentre cohort study. BJOG 2012;119:1324–1333.
Objective To assess the frequency and risk factors for surgical site infection following caesarean section.
Design Prospective multicentre cohort study.
Setting Fourteen NHS hospitals in England, April to September 2009.
Population Women who underwent caesarean section at participating hospitals during designated study periods.
Methods Infections that met standard case definitions were identified through active follow up by healthcare staff during the hospital stay, on return to hospital, during midwife home visits and through self-completed patient questionnaires.
Main outcome measure Surgical site infection within 30 days of operation.
Results Altogether, 9.6% (394/4107) of women in the study developed a postsurgical infection following caesarean section with 0.6% (23/4107) readmitted for treatment of the infection. Being overweight (body mass index [BMI] 25–30 kg/m2 odds ratio [OR] 1.6, 95% confidence interval [95% CI] 1.2–2.2) or obese (BMI 30–35 kg/m2 OR 2.4, 95% CI 1.7–3.4; BMI > 35 kg/m2 OR 3.7, 95% CI 2.6–5.2) were major independent risk factors for infection (compared with BMI 18.5–25 kg/m2). There was a suggestion that younger women, and operations performed by associate specialist and staff grade surgeons had a greater odds of developing surgical site infection with OR 1.9, 95% CI 1.1–3.4 (<20 years versus 25–30 years), and OR 1.6, 95% CI 1.0–2.4 (versus consultants), respectively.
Conclusions This study identified high rates of postsurgical infection following caesarean section. Given the number of women delivering by caesarean section in the UK, substantial costs will be incurred as a result of these infections. Prevention of these infections should be a clinical and public health priority.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
The proportion of births delivered by caesarean section in England has risen substantially over the past 30 years from 9% in 1980 to 25% in 2009/10,1 with considerable variation between English NHS Trusts, particularly in rates of emergency caesarean section.2 Postsurgical infection represents one of a number of potential complications of caesarean section3,4 with rates between 1.2 and 5.0% reported for women during their inpatient stay.5–8 Given the short length of postoperative stay in hospital, typically 2–4 days, the overall rate of infection following caesarean section is likely to be significantly higher than these estimates.5,6
Although most caesarean section wound infections are superficial, this represents a substantial burden to the health system, given the high number of women undergoing this type of surgery. However, a proportion of infections are more severe infections of the deeper tissue or reproductive organs necessitating extended hospital stays or readmission to hospital,9,10 further adding to the economic burden of these infections. On very rare occasions the development of an infection after caesarean section can lead to devastating outcomes as documented in the 2006–08 Confidential Enquiry into Maternal Deaths.11
Given the importance of these infections, and the lack of any existing national surveillance scheme following caesarean section, we developed a study to comprehensively assess the rate of surgical site infection following caesarean section through the detection of infections during the initial inpatient stay and through a number of detection methods implemented after discharge. The study also collected data on potential risk factors for these infections.
Fifteen hospitals already participating in the Health Protection Agency’s (HPA) Surgical Site Infection Surveillance Scheme agreed to take part in the caesarean section pilot study following a request for volunteers. The hospitals were all general acute hospitals, providing emergency care and general treatment for acute illness, and belonged to hospital trusts that carried out between 2100 and 11 284 deliveries a year for 2009/10, 550–2742 of which were caesarean sections.1 All 15 hospitals had a special-care baby unit, eight had a high-dependency baby unit and eight had a neonatal intensive-care unit.
Data were collected for all caesarean sections occurring at each participating hospital within designated 3-month periods to ensure that surveillance was conducted on a representative sample of women. The study took place between April and September 2009 and surveillance was conducted for either one or two 3-month periods.
The surveillance protocol and dataset were based on existing literature and agreed with advisors from the UK Royal College of Obstetricians and Gynaecologists and the Royal College of Midwives. Data were collected by designated hospital staff from theatre and hospital records on a set of demographic and operation factors including age, American Society of Anaesthesiologists (ASA) score, body mass index (BMI, kg/m2) measured at first clinic appointment, diabetes, ethnicity, gestation period, previous caesarean section, wound class, anaesthesia, anticoagulation, antimicrobial prophylaxis, blood loss and other complications, method of wound closure, operation duration and surgeon grade (junior surgeon in training: Foundation Year 1 and 2, Specialist Registrar; assistant specialist grade: Staff Grade and Associate Specialist; senior specialist: Consultant). Data were entered into the Surgical Site Infection Surveillance System database via a secure website by hospital staff.
Classification of risk factors and case definitions
The indication for caesarean section was classified according to the Royal College of Obstetricians and Gynaecologists urgency of risk categories12 as follows: immediate threat to the life of the mother or fetus, maternal or fetal compromise but no immediate threat to the life of woman or fetus, no maternal or fetal compromise but requires early delivery, or timed to suit the woman or staff.
Contamination of the surgical wound was classified using an adaptation of a standard definition13 to include membrane rupture. Hence, where membranes had ruptured less than 12 hours before caesarean section the wound was classified as clean-contaminated, and if ruptured more than 12 hours before the wound was classified as contaminated. A risk index score incorporating the modified definition of wound classification was calculated for each woman.14 This comprises an ASA score ≥3, wound classification of contaminated or dirty, and operation duration >75th centile time (60 minutes for caesarean section), with each factor contributing one point to the risk index.
The presence of surgical site infection was defined by applying a standard set of clinical and microbiological criteria (see Supplementary material, Table S1).15 The surgical site infection detected by healthcare professionals (hospital clinicians and community midwives) were categorised as i. superficial incisional surgical site infection, ii. deep incisional surgical site infection, and iii. organ/space infections (endometritis and other reproductive tract infections) adapted from the Centers for Disease Control and Prevention (CDC) definitions.16
All women whose caesarean section operation occurred within the designated 3-month surveillance periods were eligible for inclusion and formed the cohort for the study. Each caesarean section performed at a participating hospital was actively and prospectively followed up to identify infections that met the standard case definitions. Women were followed up during the inpatient stay, on readmission to hospital, on day visits to hospital and by the community midwives after discharge until the last midwife visit (a minimum of 10 days after the operation). A surveillance coordinator from each hospital was trained by the Health Protection Agency’s study team in the surveillance methodology including case definitions. Community midwives were in turn trained locally by the surveillance coordinator and provided with the case definitions and standard report forms to indicate the presence of symptoms. Infections reported by community midwives were confirmed by hospital surveillance coordinators before being entered into the database. The database provided further validation checks to ensure that symptoms met the criteria for infection.
Surgical site infections manifesting in the community and not detected elsewhere were also identified via a questionnaire completed by the woman at 30 days after the operation. Potential infections reported by the women were confirmed by the hospital surveillance coordinator using specific criteria that took into account self-reported symptoms (Table 1). Where a self-reported surgical site infection had already been detected directly by a hospital or community-based healthcare professional, the detection of the infection was ascribed to the healthcare professional.
Table 1. Criteria for identifying patient-reported surgical site infections
Indications of surgical site infection
*Clinical signs: at least two of pain, heat, redness or swelling
Discharge of pus from the wound + antibiotics
Clinical signs* + dehiscence
Clinical signs* + antibiotics
Uterine tenderness + antibiotics
Abdominal pain + antibiotics
Purulent drainage from uterus + antibiotics
Data were analysed using Microsoft Excel and STATA™ (version 11). Comparisons between groups were undertaken using chi-square and Wilcoxon rank-sum tests. Incisional and uterine infections were combined to assess risk factors for any surgical site infection that occurred within 31 days of the procedure. Risk factors were grouped according to internationally recognised categories, e.g. the World Health Organization BMI classification, or according to frequency distribution. Multivariable logistic regression analysis was undertaken using a mixed effect model to establish factors independently associated with development of a surgical site infection with allowance for variation between hospitals by including hospital identification code as a random effect in the model. The individual components of the risk index were included separately for multivariable analysis. Receiver operating characteristic (ROC) curves were used to assess the value of the final model in predicting risk of infection, with the area under the curve indicating the ability of a model to predict risk (a value of 0.5 indicates no predictive ability and a value of 1.0 indicates perfect predictive ability). The Intraclass Correlation Coefficient was calculated to determine the proportion of total variance due to variance between hospitals.
Fourteen NHS hospitals from seven Strategic Health Authorities in England completed the surveillance. One hospital withdrew after 6 weeks because they were unable to complete the surveillance on a full cohort of women through lack of resources and their consequently incomplete data were excluded from the analysis.
A total of 4107 operations were included in the study, with the number of procedures per hospital in each 3-month surveillance period ranging from 120 to 408. The median age of women included in the study was 31 years (range 14–56) and the median BMI was 25.3 (interdecile range 20.4–35.0). Ethnicity was recorded for 3883 (94.5%) women of whom 9.3% were Asian, 5.7% black and 81.8% white. The median postoperative length of hospital stay was 3 days.
Seventy-six percent (3137) of women were followed up after discharge. Post-discharge follow up to at least 10 days was achieved for 74% (3029/4107) with 36% followed up for 30 days (1482/4107). A comparison of characteristics of those followed up to 30 days using any of the detection methods, including self-completion of patient questionnaire, is shown in Table 2. Failure to follow up to 30 days was associated with a slightly younger age (30.9 versus 31.9 years), a marginally higher BMI (26.4 versus 25.8 kg/m2), a higher proportion with non-white ethnicity and not being able to read English (4.9% versus 2.1%). There were no significant differences in follow up according to the women’s diabetic status, indication for caesarean section or for number of previous caesarean sections.
Table 2. Comparison of characteristics for women followed up to 30 days postoperation by any method to those followed up for <30 days
All women (n = 4107)
Follow up 30 days (n = 1482)
Follow up <30 days (n = 2625)
*Information supplied by 12 of 14 hospitals.
Age (years) mean
z = −4.9, P < 0.01
z =3.5, P < 0.01
χ2 (4 df) = 19.9, P < 0.01
χ2 (1 df) = 0.2, P = 0.65
RCOG urgency of risk (%)
Delivery timed to suit
χ2 (3 df) = 2.68, P = 0.44
Maternal or fetal compromise
No compromise but needs early delivery
Number previous caesareans (%)
χ2 (3 df) = 4.8, P = 0.19
Unable to read English (%)*
χ2 (3 df) = 19.1, P < 0.01
Ninety-eight percent (3699/3776) of women for whom information was collected were given antimicrobial prophylaxis. In most hospitals (9/14) the first choice of antibiotic agent was co-amoxiclav. A further four hospitals used cefuroxime and one used a combination of cefradine and metronidazole. The timing of administration of the antibiotic varied between hospitals with eight hospitals administering antibiotics after delivery of the baby, of which three specified that this was after cord clamping. A further three hospitals administered the antibiotic at induction of anaesthesia and three gave it during the operation.
Surgical site infections detected
Applying all methods of detection, a total of 394 surgical site infections were identified from 4107 operations, representing a risk of 9.6%. Three of the women were reported to have developed both an incisional and a uterine infection. Eleven percent (44/394) of infections were detected during the initial admission (21) or through readmission to hospital (23), 55% (218/394) were detected post-discharge by community midwife or other hospital-based healthcare professional and 34% (132/394) were reported by the woman and confirmed by the surveillance coordinator. In all, 95% of the infections were identified after the initial inpatient stay.
Of the 394 infections, 348 (88.3%) were superficial incisional, 19 (4.8%) were deep incisional and 27 (6.9%) were organ/space infections, 25 of which were classed as endometritis. The median time to infection for all surgical site infections was 10 days and for deep and organ/space infections alone was 8 days. Of the 4107 women followed up by the study, 0.6% (23/394) were readmitted to hospital for treatment of their infection.
Causative microorganisms were recorded for 39.8% of the infections (157/394). Of these infections 24.2% (38/157) were reported to be polymicrobial. The most commonly reported pathogen was Staphylococcus aureus (40.4%) of which 17.1% were methicillin-resistant. Other pathogens included anaerobic cocci (23.2%), Enterobacteriaceae (13.3%) and streptococci (7.4%) as shown in Table 3.
Table 3. Microorganisms reported as causing surgical site infection following caesarean section in a multicentre study in England (2009)
*Causative microorganisms were not reported for 60% of infections (237 of 394 surgical site infections).
Methicillin-resistant S. aureus
Methicillin-susceptible S. aureus
Anaerobic cocci (unspecified)
Other coliforms (unspecified)
Streptococcus agalactiae (group B)
Streptococcus pyogenes (group A)
Streptococcus—other aerobic sp.
Single variable analysis indicated that a number of factors were associated with the development of an infection (Tables 4 and 5) including BMI, diabetes, surgeon grade and ASA score. Obesity was strongly associated with development of a surgical site infection, with risk increasing with each successive category of BMI (Figure 1). Separate single variable analyses of the association of BMI with either superficial incisional or deep incisional and organ/space infections indicated that BMI was significantly associated with development of both superficial (P < 0.0001) and deep/organ space (P < 0.003) infections.
Table 4. Patient-related risk factors for surgical site infection following caesarean section in England (2009)
Patient-related risk factor
Infection rate (%)
Unadjusted OR (95% CI)
Adjusted OR* (95% CI)
χ2; (df), P
*With adjustment for variation between hospitals; also adjusted for operation duration and surgeon category (Table 5).
Age in years (n = 4105)
9.35 (6), 0.15
ASA score (n = 3128)
BMI category (n = 3910)
65.48 (4), <0.01
Category of risk (n = 4013)
Delivery timed to suit
Maternal or fetal compromise
No compromise but needs early delivery
Diabetes (n = 3917)
Ethnicity (n = 3883)
4.10 (4), 0.39
Pregnancy gestation (n = 3976)
Previous caesarean section (n = 3766)
Table 5. Operation-related risk factors for surgical site infection following caesarean section in England (2009)
Operation-related risk factor
Infection rate (%)
Unadjusted OR (95% CI)
Adjusted OR* (95% CI)
χ2; (df), P
*With adjustment for variation between hospitals; also adjusted for age, BMI, ethnicity (Table 4).
Anaesthesia (n = 3962)
More than one method
Anticoagulation (n = 3087)
Antimicrobial prophylaxis (n = 3776)
Blood loss in ml (n = 3963)
Complicated caesarean section (n = 3740)
Extension uterine angles/vaginal tear
Fat closure (n = 3067)
Method of closure (n = 3935)
Operation duration (n = 3902)
3.47 (3), 0.32
Surgeon grade (n = 3923)
5.66 (4), 0.23
Staff and Assistant Specialist grade
Specialist Registrar (trainee)
FY1 and FY2
Wound class (n = 4082)
3 and 4
Risk index (n = 3001)
Diabetic status was reported for 3917 women, of whom 5.6% (218) had either gestational diabetes (160; 4.1%) or pre-existing diabetes type I (41, 1.0%) or type II (17; 0.4%). The risk of surgical site infection for women with diabetes was 15.6% (95% CI 11.0–21.1%) compared with 9.6% (95% CI 8.7–10.6%) for non-diabetic women (odds ratio 1.8, 95% CI 1.2–2.6). Operations where the lead surgeon was an associate specialist or staff grade were associated with a higher likelihood of infection than those performed by consultants (13.1%, 95% CI 10.3–16.3% versus 7.9%, 95% CI 6.0–10.2%; odds ratio 1.8, 95% CI 1.2–2.6). Most procedures (61%) were performed by specialist registrars. Women with an ASA score of 2 also had a significantly higher risk of infection than those with a score of 1 (odds ratio 1.5, 95% CI 1.2–2.0). There was no significant difference in risk of surgical site infection associated with either the risk index score or Royal College of Obstetricians and Gynaecologists categories of risk.
The surgical site infection rate showed a broadly U-shaped distribution for women’s age; higher in women <20 years (13%, 95% CI 7.9–19.3%) and women over 45 years (16%, 95% CI 5.3–32.8%; Table 4) compared with the 25–30-year age group (9.2%, 95% CI 7.4–11.2%) but was not significant in the single variable analysis.
Multivariable analysis generated a model in which factors were retained if they were significant as a whole, or were significant (or approached significance) in one of the individual strata (Tables 4 and 5). Following adjustment for the other patient and operation-related risk factors in the multivariable analysis, including variation between hospitals, only BMI was found to be significantly associated with developing a surgical site infection. Compared with women with a normal BMI (18.5–25 kg/m2) overweight women (BMI 25–30) were found to have an estimated odds of infection 1.6 (95% CI 1.2–2.2) times greater than normal weight women, and obese women (BMI > 30) 2.4 times greater (95% CI 1.7–3.4). Given the potential for effect modification between BMI and other variables in the model, these were tested for but not found to be significant.
Although age and surgeon grade as a whole were not significantly associated with developing a surgical site infection in the multivariable model, specific categories within these variables showed evidence of an increased odds of infection. There was some evidence of an increased likelihood of infection in women aged <20 years (odds ratio 1.9, 95% CI 1.1–3.4) compared with women aged 25–30 years, and operations performed by associate specialist and staff grade surgeons (odds ratio 1.6, 95% CI 1.0–2.4) compared with consultants. Neither ethnicity nor duration of operation were significantly associated with surgical site infection but there was weak evidence that women of black ethnicity and operations of 55–70 minutes were associated with increased odds of infection.
Given that only 36% of women had documented follow up for the entire 30-day period, whereas 74% had documented follow up for at least 10 days, multivariable analyses were repeated for infections detected within the first 10 days (55% of all infections). This confirmed BMI (P < 0.0001) and young age (age <20 years, P = 0.04) as risk factors for infection with no other factors reaching statistical significance in this reduced dataset.
The area under the ROC curves for predicting the risk of surgical site infection following caesarean section based on BMI, age, ethnicity, surgeon grade and operation duration was 0.64. This compared with a lower value of 0.52 for the ROC curve based on the risk index with adapted wound class, which indicates minimal predictive power.
The Intraclass Correlation Coefficient was 0.06 (P < 0.01) indicating that only 6% of the total variance in the occurrence of infections was accounted for by between-hospital variance.
Results from this study have identified high rates of surgical site infection following caesarean section, with one in ten operations leading to an infection. This is substantially higher than rates of infection for abdominal hysterectomy (6.6%) and approaches rates associated with large bowel surgery (12.7%) identified using equivalent detection methods (Wloch 2011, personal observation).
The hospitals taking part in the study covered both urban and rural catchment areas and included small and large maternity units. As there are approximately 160 000 caesarean sections performed each year in England,1 if the findings from this study were extrapolated nationally this would suggest over 15 000 women undergoing caesarean section develop postsurgical infections each year, with 960 (0.6%) readmitted as a result of their infection. Surgical site infection occurring during the initial inpatient stay or resulting in readmission will incur significant costs to the hospital. However, the majority of costs resulting from these infections will be borne by community healthcare as a result of extended community midwife care, visits to general practitioners and antibiotic prescribing. There is currently little information available on the effect of surgical site infection on the woman’s experience and quality of life although undoubtedly it will have some impact on family life.
This study found that prophylactic antibiotics were administered to nearly all women for whom information was recorded. The Department of Health’s High Impact Intervention Care Bundle to prevent surgical site infection recommends that antibiotic prophylaxis is given within 60 minutes before incision.17 As a consequence of concerns regarding antibiotics crossing the placenta and possibly altering the baby’s gut flora, affecting development of the immune system or masking neonatal sepsis,18 several hospitals administer antibiotics after cord clamping to minimise fetal exposure. However, recent debate suggests that pre-incision broad-spectrum antibiotics lower rates of maternal infection without demonstrable short-term effects on neonatal outcomes.19 Further investigation of the impact of timing of administration on specific neonatal outcomes is needed.
Most causative organisms reported were common skin or female genital tract flora. The β-lactams employed for antimicrobial prophylaxis in many of the hospitals in this study are suitable for targeting such organisms, which suggests that current prophylactic regimens are appropriate.
An increased risk of surgical site infection in younger women has been reported in two studies5,20 but another study in Scotland found the converse, that risk of infection increased slightly with increasing age.6 The data from this study suggest that younger women (<20 years) have high surgical site infection rates with significantly increased odds of infection when adjusted for other surgical and patient risk factors. The reasons for this increase in young women are unclear and worthy of further consideration because this study found the increase was unlikely to be the result of ascertainment bias because there was no difference in length of hospital stay or in days of midwife follow up between young women (<20 years) and older women.
An association between surgical site infection and BMI has been noted in several studies.5,6,21,22 This may be a result of the standard dosage of prophylactic antibiotics achieving inadequate tissue concentrations in obese women,23 particularly if there has been no adjustment to the dose in proportion to BMI. In addition, relatively poor perfusion of adipose tissue may impair wound healing, decrease the local immune response and so facilitate infection becoming established.24–26 The incision for obese women may also need to be longer and therefore involve more tissue becoming exposed to contamination.23 Although suturing of the subcutaneous tissue space is recommended to reduce the risk of wound infection and dehiscence for women with >2 cm of subcutaneous fat, as yet there is no recommendation on the optimal incision type for obese women.27 However, there may be limited opportunities to tackle obesity by targeting dietary advice in early pregnancy or preconception.27
Although this study included more than 4000 women, it may not be sufficiently large to provide a conclusive answer to whether the increased incidence of infection in operations performed by associate specialist and staff grade surgeons was a real effect. It is also possible that this finding was the result of confounding caused by some other characteristic of the women undergoing caesarean section that was not adjusted for in the model. The lack of significance of diabetic status in the multivariable analysis is probably a result of the adjustment by BMI category.
The standard risk index for predicting risk of postsurgical infection is of limited use for caesarean section as it does not clearly stratify risk in this group of women.28 This is borne out by this study where even the modification of wound classification to incorporate risk from rupture of membranes only enabled the risk index to achieve a ROC curve of 0.52, whereas the model derived from the multivariable analysis in this study based on BMI, age, ethnic group, surgeon grade and operation duration achieved a ROC score of 0.64. Although this score is low for predictive purposes using this model or one of the procedure-specific risk models suggested by other studies5,20,29,30 may be more appropriate than the standard risk index when comparing rates of surgical site infection following caesarean section delivery.
The majority of infections reported from caesarean sections are superficial incisional infections. There were too few deep incisional or organ/space (uterine and reproductive tract infections) to undertake a separate multivariable analysis with deep and organ/space infections as outcomes but single variable logistic regression of BMI indicated that BMI remained a significant risk factor for deep and organ/space infections. This suggests that strategies aimed at controlling obesity would have an impact on all types of surgical site infection. Further studies are warranted on a more comprehensive group of women however, given that some authors propose that the risk factors for endometritis and incisional infections may be different.20,31,32
Although superficial incisional infections are relatively minor, they are still likely to result in pain and discomfort, require antimicrobial therapy and may progress to affect deeper tissues. Avoiding more severe infections is a priority, but monitoring superficial as well as deep infections provides a higher sensitivity with which to examine the quality of care and detect potential problems with infection prevention. Furthermore, all infections will result in costs to the health service through extended visits by community midwives and visits to general practitioners.
Although there remains the possibility of some variation between hospitals in the detection and reporting of wound infections, steps were taken to ensure that surgical site infections were reported as objectively and consistently as possible. Hospital staff and community midwives were trained to apply standard case definitions. Infections were confirmed by the surveillance coordinator and further validation checks were applied by the web-based data entry software to verify that symptoms met the criteria for infection.
Post-discharge follow-up data were not obtained for all women although each woman had an opportunity to report any infection that developed after their midwife care had ceased. For the majority of women data on the occurrence of surgical site infection was available at 10 days after surgery, but these data were only available for a third of women at 30 days after surgery. Undoubtedly this would have resulted in an under-ascertainment of incisional infections manifesting after discharge, and an underestimate of the true incidence of surgical site infection. However, hospitals did continue to monitor for women readmitted or visiting outpatient clinics with a surgical site infection up to 30 days, so serious deep and organ/space infections will have been better ascertained. There was little evidence of disparity between women followed up for 30 days and those with <30 days of follow up. There were minor differences in BMI and age but more marked differences were observed in ethnicity and ability to read English. Clearly the latter factor will have a direct bearing on the woman’s ability to complete the patient questionnaire unaided and therefore on our detection of infections in these women. Provision of copies of the patient questionnaire in other languages would help to reduce this bias. However, the analysis of infections detected up to 10 days after the operation found that the outcome was broadly similar to the main analysis, indicating no major evidence of bias caused by variation in the period of follow up.
This study identified high rates of postsurgical wound infection following caesarean section; considering the large number of women undergoing caesarean section in the UK, these infections are likely to incur substantial additional healthcare costs. Significant independent risk factors for wound infection were obesity, age <20 years and operations performed by associate specialist or staff grade surgeons. Although the age of women presenting for caesarean section is not amenable to clinical intervention, dietary advice, optimising surgical technique and identification of the most appropriate dosing of prophylactic antibiotics could provide a means for reducing wound infections in obese women. Given the increasing prevalence of obesity in women of childbearing age, from one in seven in 1997 to one in five in 2008,33 identification of effective measures to improve wound healing and prevent infection after caesarean section is becoming increasingly urgent.
Disclosure of interests
All authors report no potential conflicts of interest.
Contribution to authorship
All authors participated in the editing of this manuscript and approved the final version for publication. JW conceived and designed the study, with input from CW, PH and Christine McDougall. CW coordinated the study and undertook the analyses. CW, JW and TL drafted the paper. ES provided microbiological expertise and AC provided statistical advice. TL is the guarantor.
Details of ethics approval
The Health Protection Agency has National Information Governance Board for Health and Social Care approval for the collation of surveillance data in accordance with section 251 of the NHS Act 2006. No additional ethical approval was required to undertake this study.
The study received no external funding.
We thank Suzanne Elgohari for advice on analysis, Kalpna Pindoria for administrative assistance, and Andrew Hall for critical review. We would like to thank the staff in the Infection Control and Maternity Departments at the participating hospitals and the community midwives who were involved in the study: Dorset County NHS Foundation Trust, Shirley Pike, Rosie Limbach and Melanie Hunter (Maternity); George Eliot NHS Trust, Karen Shorthose and Dawn Fuller (Maternity); James Paget University Hospitals NHS Foundation Trust, Linda Hawtin and Kati Rowse-Smith (Infection Prevention & Control); Kingston Hospital NHS Trust, Louise Doyle (Maternity), Victoria Wells (Infection Prevention and Control); Maidstone and Tunbridge Wells NHS Trust, Helen Gregson (Infection Prevention); Plymouth Hospitals NHS Trust; Royal Berkshire Hospital NHS Foundation Trust, Bernadeth Rafanan, Nilangi Virgincar (Infection Prevention & Control) and Gill Valentine (Obstetrics & Gynaecology); Royal Devon and Exeter NHS Foundation Trust; The Royal Wolverhampton Hospitals NHS Trust; St Helens and Knowsley Hospitals NHS Trust; South Devon Health Care NHS Trust; The Whittington Hospital NHS Trust.